This invention relates generally to personal computers, workstations, and servers. More specifically, this invention relates to thermal and acoustic control of personal computer platforms, workstation platforms, and server platforms (xe2x80x9cplatformsxe2x80x9d or xe2x80x9csystemsxe2x80x9d). This invention further relates to a processor arrangement and processor thermal interface within a computer platform.
A major concern with personal computer, workstation, and server platforms is proper control of the thermal characteristics inside the platform""s housing. Today""s computer platforms generally have only a single, unitary housing, such as a desktop casing, mini-tower casing, or tower casing, into which all of the computer""s electronic components are placed. Inside these casings, most, if not all, of the electronic components are located within a single thermal environment.
Almost all electronic components generate heat when operated. And, different components produce different amounts of heat. The amount of heat an electronic component generates typically depends on its power consumption, and, therefore, even a single component produces different amounts of heat at different times, depending on the amount of work it is doing. Microprocessors (also called xe2x80x9cprocessorsxe2x80x9d or xe2x80x9cCPUsxe2x80x9d), in particular, generate varying amounts of heat during operation, depending on the amount of information they are required to process.
If an electronic component overheats, malfunction or premature failure of the component is likely to result. Cooling the electronic components is therefore necessary. At the same time, however, overcooling may result in condensation of water vapor within the computer housing. Liquid and refrigeration cooling methods, in particular, are prone to creating condensation problems inside conventional computer platform housings. In addition, because different electronic components generate different amounts of heat, different areas within the computer housing have different cooling requirements. The cooling requirements of components are often in contention with one another, including those of components in close proximity properly controlling the thermal characteristics of conventional platforms is difficult.
The conventional approach to thermal control of computer platforms has included mainly point solutions. Point solutions focus on the components that produce the greatest amount of heat, such as the processor, or on specific areas within the computer housing that have the greatest need for temperature control. Unfortunately, because temperature exchange within conventional computer housings is generally uncontrolled, this point approach can result in some components being cooled insufficiently while other components become overcooled. The point solution approach consequently results in thermal chaos within the computer housing.
Another problem in the industry is that modern computer systems are not easy to adapt or to upgrade. Replacing worn or antiquated components can be difficult. Because of the rapid advance in microprocessor and other computing technologies, many consumers desire the ability to upgrade their computer systems by replacing individual components of the system, rather than replacing the system entirely.
Upgrading a conventional computer system generally requires opening the casing, identifying the parts for replacement, disconnecting them, removing them, and putting in the new components. This can require the user to disconnect and reconnect power supply or other cabling, among other things. Upgrading a computer platform often requires a higher degree of computer expertise than many consumers have. Furthermore, upgraded components often have higher cooling requirements than the components they replace. This is typically the case with microprocessors, because, as processing speeds increase, the heat output and cooling requirements of the processors increase correspondingly. Cooling systems, therefore, may need to be upgraded when other components are upgraded.
Yet another significant problem in the industry is meeting the acoustic goals, or xe2x80x9cquiet officexe2x80x9d requirements, of the computer platform. Consumers desire computer platforms that do not generate excessive noise during operation. A conventional xe2x80x9cquiet officexe2x80x9d requirement for a computer platform, for instance, requires that noise from the computer system be less than approximately 5.2 bels. The industry has had difficulty producing systems that meet both the cooling needs of the platform and these xe2x80x9cquiet officexe2x80x9d requirements. This is because the primary noise generating components in a computer platform are the cooling systems, including the CPU fan(s) and the system fan(s), in particular. Other major contributors to the noise of a system include the power supply and the drives.
Unfortunately, most conventional systems are generally inefficient in delivering cooling air from the fans to the components to be cooled. Furthermore, conventional cooling techniques cool based on a volume of air (or xe2x80x9cthroughput,xe2x80x9d measured in terms of cubic feet per minute (CFM)) supplied across the components. To increase the cooling performance of the system, therefore, the air throughput must be increased. This can be done, for example, by increasing the operating speed of the fans or the size of the fans. Either of these two approaches generally results in noisier cooling systems and, hence, noisier platform operation.
Ducting has been attempted as a way of more efficiently directing air from the cooling fans to the components to be cooled. Ducting alone, however, has not proven efficient in meeting the cooling requirements of a system. Ducting""s inefficiency is due in part to significant air leakage (i.e., around 35%) from the ducting because of the difficulty in fitting the duct around computer components. As noted above, inefficient cooling makes it difficult for the platform to meet both cooling and acoustic goals.
Yet another problem in prior art computer platforms is the complexity of the processor arrangement and its cooling interface. Conventional processor arrangements, particularly for multiple processor systems, are bulky and structurally complex. Specifically, prior art processor arrangements typically use a board-mounted retention scheme to support the processors, the processor heatsinks, the power pods, and the power pod heatsinks. Unfortunately, the prior art arrangement occupies more board area than desirable and is also volumetrically inefficient.
Additionally, board-mounting raises additional concerns relating to the strength of the board, the location of board supports, and shock and vibration concerns, for example. In particular, shock and/or vibration of the computer platform can seriously jeopardize the structural integrity of a CPU board that uses a board-mounted retention scheme. This is because shock and/or vibration causes the heatsinks for both the processors and the power pods to move. Because the heatsinks are heavy, their movement causes stress in the CPU board. CPU board stress can seriously damage the CPU board and/or board-mounted components. For these reasons, the allowable size and mass of the power pod and CPU heatsinks are limited in such an arrangement. Board mounting can also create tolerance stack-up problems because the board-mounted components are typically located directly on top of each other on the CPU board, causing their tolerances to add up.